MathScape
收藏资源简介:
MathScape是由南开大学、北京大学、百川公司和中国科学院大学联合创建的多模态数学问题数据集,旨在评估多模态大型语言模型在数学问题解决中的应用。该数据集包含1325张图像,涵盖从小学到高中的数学问题,难度从易到难,涉及多种题型和知识领域。数据集的创建过程包括将数学文档转换为图像、拍摄照片和屏幕截图,以及进行质量检查和知识分类。MathScape数据集主要用于评估模型在复杂数学问题解决中的表现,特别是在结合视觉和文本信息的能力。
MathScape is a multimodal mathematical problem dataset jointly created by Nankai University, Peking University, Baichuan Company, and the University of Chinese Academy of Sciences, aiming to evaluate the application of multimodal large language models in mathematical problem solving. This dataset contains 1325 images, covering mathematical problems from primary school to high school with difficulties ranging from easy to hard, involving diverse question types and knowledge domains. The construction process of the dataset includes converting mathematical documents into images, collecting photos and screenshots, as well as conducting quality checks and knowledge categorization. The MathScape dataset is primarily used to evaluate models' performance in solving complex mathematical problems, especially their ability to integrate visual and textual information.




